Methodology and notes - AI Skills and Adoption report

This report draws on a survey of BCS members conducted online between 25 June and 30 July 2026. A total of 844 BCS members took part in the survey, which was promoted to the membership via email and newsletter.

Respondents work across the public sector, private sector and academia, and in organisations of every size. By role, the largest groups were IT professionals and IT leaders, alongside retired members, students and others. The great majority are based in the United Kingdom.

Percentages are based on the number of respondents who answered each question, which varies from question to question because those who skipped a question are excluded from its base. Where a question asked respondents to rank their top three choices, a figure such as ‘75% placed upskilling in their top three’ means that that proportion of those who answered ranked the option first, second or third. Some percentages may not sum to 100 due to rounding or, for ‘select all' and ‘top three’ questions, because respondents could choose more than one option.

Open-text responses on AI skills were coded thematically against a framework of skill categories developed from the responses themselves. A single response could be coded to more than one category, because respondents were asked for up to two examples, so percentages sum to more than 100. Percentages are based on the 530 substantive responses, excluding those who answered that AI was not relevant to their organisation or who gave no usable answer. Around a fifth of substantive responses could not be assigned to a category and are excluded from the figures shown. This coding is indicative of the relative weight of each theme rather than a precise measurement, and the findings are best read as a rank order. Illustrative quotations are reproduced as written.

Where the report compares findings with BCS's Tech priorities, skills and the AI outlook for 2025, the comparison is one of direction only: the two surveys asked related but not identical questions of different respondent groups, so figures are not directly comparable.

The coding, which assigned each response to one or more skill categories, was carried out with assistance from AI tools and then checked by BCS staff against a random 10% sample, which showed 86% agreement. 

The rating questions asked respondents to score from 1 to 5, but the survey tool also allowed a response of zero, which 16 respondents chose for each question. These responses have been kept, as they most likely indicate the lowest possible rating. Excluding them would raise the averages slightly, to 2.92 for adoption effectiveness and 3.37 for ethical and responsible use, and would not change the findings. 

The findings, analysis and arguments in this report were developed by BCS staff, who worked through the responses to every survey question and recorded detailed notes and commentary on each. AI tools were used to help analyse the survey data and to organise, format and edit those notes into a report. All conclusions and recommendations were determined by BCS staff, all survey data and quotations come directly from respondents, and all survey figures were checked against the underlying survey data.